author | wenzelm |
Fri, 18 Mar 2016 16:26:35 +0100 | |
changeset 62663 | bea354f6ff21 |
parent 61707 | d5ddd022a451 |
child 62819 | d3ff367a16a0 |
permissions | -rw-r--r-- |
44698 | 1 |
(* Title: Pure/PIDE/xml.ML |
2 |
Author: David Aspinall |
|
3 |
Author: Stefan Berghofer |
|
4 |
Author: Makarius |
|
24264 | 5 |
|
46840 | 6 |
Untyped XML trees and representation of ML values. |
24264 | 7 |
*) |
8 |
||
43767 | 9 |
signature XML_DATA_OPS = |
10 |
sig |
|
43778
ce9189450447
more compact representation of XML data (notably sort/typ/term), using properties as vector of atomic values;
wenzelm
parents:
43768
diff
changeset
|
11 |
type 'a A |
43767 | 12 |
type 'a T |
43778
ce9189450447
more compact representation of XML data (notably sort/typ/term), using properties as vector of atomic values;
wenzelm
parents:
43768
diff
changeset
|
13 |
type 'a V |
ce9189450447
more compact representation of XML data (notably sort/typ/term), using properties as vector of atomic values;
wenzelm
parents:
43768
diff
changeset
|
14 |
val int_atom: int A |
ce9189450447
more compact representation of XML data (notably sort/typ/term), using properties as vector of atomic values;
wenzelm
parents:
43768
diff
changeset
|
15 |
val bool_atom: bool A |
ce9189450447
more compact representation of XML data (notably sort/typ/term), using properties as vector of atomic values;
wenzelm
parents:
43768
diff
changeset
|
16 |
val unit_atom: unit A |
43767 | 17 |
val properties: Properties.T T |
18 |
val string: string T |
|
19 |
val int: int T |
|
20 |
val bool: bool T |
|
21 |
val unit: unit T |
|
22 |
val pair: 'a T -> 'b T -> ('a * 'b) T |
|
23 |
val triple: 'a T -> 'b T -> 'c T -> ('a * 'b * 'c) T |
|
24 |
val list: 'a T -> 'a list T |
|
25 |
val option: 'a T -> 'a option T |
|
43778
ce9189450447
more compact representation of XML data (notably sort/typ/term), using properties as vector of atomic values;
wenzelm
parents:
43768
diff
changeset
|
26 |
val variant: 'a V list -> 'a T |
43767 | 27 |
end; |
28 |
||
24264 | 29 |
signature XML = |
30 |
sig |
|
46837
5bdd68f380b3
clarified XML signature (again) -- coincide with basic Markup without explicit dependency;
wenzelm
parents:
45155
diff
changeset
|
31 |
type attributes = (string * string) list |
24264 | 32 |
datatype tree = |
46837
5bdd68f380b3
clarified XML signature (again) -- coincide with basic Markup without explicit dependency;
wenzelm
parents:
45155
diff
changeset
|
33 |
Elem of (string * attributes) * tree list |
24264 | 34 |
| Text of string |
38266
492d377ecfe2
type XML.body as basic data representation language;
wenzelm
parents:
38228
diff
changeset
|
35 |
type body = tree list |
49650
9fad6480300d
support for wrapped XML elements, which allows to preserve full markup tree information in to_XML/from_XML conversion;
wenzelm
parents:
49599
diff
changeset
|
36 |
val wrap_elem: ((string * attributes) * tree list) * tree list -> tree |
9fad6480300d
support for wrapped XML elements, which allows to preserve full markup tree information in to_XML/from_XML conversion;
wenzelm
parents:
49599
diff
changeset
|
37 |
val unwrap_elem: tree -> (((string * attributes) * tree list) * tree list) option |
26546 | 38 |
val add_content: tree -> Buffer.T -> Buffer.T |
39555
ccb223a4d49c
added XML.content_of convenience -- cover XML.body, which is the general situation;
wenzelm
parents:
38474
diff
changeset
|
39 |
val content_of: body -> string |
56059
2390391584c2
some document antiquotations for Isabelle/jEdit elements;
wenzelm
parents:
49650
diff
changeset
|
40 |
val trim_blanks: body -> body |
26546 | 41 |
val header: string |
42 |
val text: string -> string |
|
43 |
val element: string -> attributes -> string list -> string |
|
40131
7cbebd636e79
explicitly qualify type Output.output, which is a slightly odd internal feature;
wenzelm
parents:
39555
diff
changeset
|
44 |
val output_markup: Markup.T -> Output.output * Output.output |
26539
a0754be538ab
added output_markup (from Tools/isabelle_process.ML);
wenzelm
parents:
26525
diff
changeset
|
45 |
val string_of: tree -> string |
43791 | 46 |
val pretty: int -> tree -> Pretty.T |
60982
67e389f67073
precise BinIO, without newline conversion on Windows;
wenzelm
parents:
58854
diff
changeset
|
47 |
val output: tree -> BinIO.outstream -> unit |
26984
d0e098e206f3
added parse_document (optional unchecked header material);
wenzelm
parents:
26554
diff
changeset
|
48 |
val parse_comments: string list -> unit * string list |
24264 | 49 |
val parse_string : string -> string option |
26546 | 50 |
val parse_element: string list -> tree * string list |
26984
d0e098e206f3
added parse_document (optional unchecked header material);
wenzelm
parents:
26554
diff
changeset
|
51 |
val parse_document: string list -> tree * string list |
26539
a0754be538ab
added output_markup (from Tools/isabelle_process.ML);
wenzelm
parents:
26525
diff
changeset
|
52 |
val parse: string -> tree |
43767 | 53 |
exception XML_ATOM of string |
54 |
exception XML_BODY of body |
|
43778
ce9189450447
more compact representation of XML data (notably sort/typ/term), using properties as vector of atomic values;
wenzelm
parents:
43768
diff
changeset
|
55 |
structure Encode: XML_DATA_OPS |
ce9189450447
more compact representation of XML data (notably sort/typ/term), using properties as vector of atomic values;
wenzelm
parents:
43768
diff
changeset
|
56 |
structure Decode: XML_DATA_OPS |
24264 | 57 |
end; |
58 |
||
59 |
structure XML: XML = |
|
60 |
struct |
|
61 |
||
26546 | 62 |
(** XML trees **) |
63 |
||
46837
5bdd68f380b3
clarified XML signature (again) -- coincide with basic Markup without explicit dependency;
wenzelm
parents:
45155
diff
changeset
|
64 |
type attributes = (string * string) list; |
26546 | 65 |
|
66 |
datatype tree = |
|
46837
5bdd68f380b3
clarified XML signature (again) -- coincide with basic Markup without explicit dependency;
wenzelm
parents:
45155
diff
changeset
|
67 |
Elem of (string * attributes) * tree list |
28033 | 68 |
| Text of string; |
26546 | 69 |
|
38266
492d377ecfe2
type XML.body as basic data representation language;
wenzelm
parents:
38228
diff
changeset
|
70 |
type body = tree list; |
492d377ecfe2
type XML.body as basic data representation language;
wenzelm
parents:
38228
diff
changeset
|
71 |
|
49650
9fad6480300d
support for wrapped XML elements, which allows to preserve full markup tree information in to_XML/from_XML conversion;
wenzelm
parents:
49599
diff
changeset
|
72 |
|
9fad6480300d
support for wrapped XML elements, which allows to preserve full markup tree information in to_XML/from_XML conversion;
wenzelm
parents:
49599
diff
changeset
|
73 |
(* wrapped elements *) |
9fad6480300d
support for wrapped XML elements, which allows to preserve full markup tree information in to_XML/from_XML conversion;
wenzelm
parents:
49599
diff
changeset
|
74 |
|
9fad6480300d
support for wrapped XML elements, which allows to preserve full markup tree information in to_XML/from_XML conversion;
wenzelm
parents:
49599
diff
changeset
|
75 |
val xml_elemN = "xml_elem"; |
9fad6480300d
support for wrapped XML elements, which allows to preserve full markup tree information in to_XML/from_XML conversion;
wenzelm
parents:
49599
diff
changeset
|
76 |
val xml_nameN = "xml_name"; |
9fad6480300d
support for wrapped XML elements, which allows to preserve full markup tree information in to_XML/from_XML conversion;
wenzelm
parents:
49599
diff
changeset
|
77 |
val xml_bodyN = "xml_body"; |
9fad6480300d
support for wrapped XML elements, which allows to preserve full markup tree information in to_XML/from_XML conversion;
wenzelm
parents:
49599
diff
changeset
|
78 |
|
9fad6480300d
support for wrapped XML elements, which allows to preserve full markup tree information in to_XML/from_XML conversion;
wenzelm
parents:
49599
diff
changeset
|
79 |
fun wrap_elem (((a, atts), body1), body2) = |
9fad6480300d
support for wrapped XML elements, which allows to preserve full markup tree information in to_XML/from_XML conversion;
wenzelm
parents:
49599
diff
changeset
|
80 |
Elem ((xml_elemN, (xml_nameN, a) :: atts), Elem ((xml_bodyN, []), body1) :: body2); |
9fad6480300d
support for wrapped XML elements, which allows to preserve full markup tree information in to_XML/from_XML conversion;
wenzelm
parents:
49599
diff
changeset
|
81 |
|
9fad6480300d
support for wrapped XML elements, which allows to preserve full markup tree information in to_XML/from_XML conversion;
wenzelm
parents:
49599
diff
changeset
|
82 |
fun unwrap_elem (Elem ((name, (n, a) :: atts), Elem ((name', atts'), body1) :: body2)) = |
9fad6480300d
support for wrapped XML elements, which allows to preserve full markup tree information in to_XML/from_XML conversion;
wenzelm
parents:
49599
diff
changeset
|
83 |
if name = xml_elemN andalso n = xml_nameN andalso name' = xml_bodyN andalso null atts' |
9fad6480300d
support for wrapped XML elements, which allows to preserve full markup tree information in to_XML/from_XML conversion;
wenzelm
parents:
49599
diff
changeset
|
84 |
then SOME (((a, atts), body1), body2) else NONE |
9fad6480300d
support for wrapped XML elements, which allows to preserve full markup tree information in to_XML/from_XML conversion;
wenzelm
parents:
49599
diff
changeset
|
85 |
| unwrap_elem _ = NONE; |
9fad6480300d
support for wrapped XML elements, which allows to preserve full markup tree information in to_XML/from_XML conversion;
wenzelm
parents:
49599
diff
changeset
|
86 |
|
9fad6480300d
support for wrapped XML elements, which allows to preserve full markup tree information in to_XML/from_XML conversion;
wenzelm
parents:
49599
diff
changeset
|
87 |
|
9fad6480300d
support for wrapped XML elements, which allows to preserve full markup tree information in to_XML/from_XML conversion;
wenzelm
parents:
49599
diff
changeset
|
88 |
(* text context *) |
9fad6480300d
support for wrapped XML elements, which allows to preserve full markup tree information in to_XML/from_XML conversion;
wenzelm
parents:
49599
diff
changeset
|
89 |
|
9fad6480300d
support for wrapped XML elements, which allows to preserve full markup tree information in to_XML/from_XML conversion;
wenzelm
parents:
49599
diff
changeset
|
90 |
fun add_content tree = |
9fad6480300d
support for wrapped XML elements, which allows to preserve full markup tree information in to_XML/from_XML conversion;
wenzelm
parents:
49599
diff
changeset
|
91 |
(case unwrap_elem tree of |
9fad6480300d
support for wrapped XML elements, which allows to preserve full markup tree information in to_XML/from_XML conversion;
wenzelm
parents:
49599
diff
changeset
|
92 |
SOME (_, ts) => fold add_content ts |
9fad6480300d
support for wrapped XML elements, which allows to preserve full markup tree information in to_XML/from_XML conversion;
wenzelm
parents:
49599
diff
changeset
|
93 |
| NONE => |
9fad6480300d
support for wrapped XML elements, which allows to preserve full markup tree information in to_XML/from_XML conversion;
wenzelm
parents:
49599
diff
changeset
|
94 |
(case tree of |
9fad6480300d
support for wrapped XML elements, which allows to preserve full markup tree information in to_XML/from_XML conversion;
wenzelm
parents:
49599
diff
changeset
|
95 |
Elem (_, ts) => fold add_content ts |
9fad6480300d
support for wrapped XML elements, which allows to preserve full markup tree information in to_XML/from_XML conversion;
wenzelm
parents:
49599
diff
changeset
|
96 |
| Text s => Buffer.add s)); |
26546 | 97 |
|
39555
ccb223a4d49c
added XML.content_of convenience -- cover XML.body, which is the general situation;
wenzelm
parents:
38474
diff
changeset
|
98 |
fun content_of body = Buffer.empty |> fold add_content body |> Buffer.content; |
ccb223a4d49c
added XML.content_of convenience -- cover XML.body, which is the general situation;
wenzelm
parents:
38474
diff
changeset
|
99 |
|
26546 | 100 |
|
56059
2390391584c2
some document antiquotations for Isabelle/jEdit elements;
wenzelm
parents:
49650
diff
changeset
|
101 |
(* trim blanks *) |
2390391584c2
some document antiquotations for Isabelle/jEdit elements;
wenzelm
parents:
49650
diff
changeset
|
102 |
|
2390391584c2
some document antiquotations for Isabelle/jEdit elements;
wenzelm
parents:
49650
diff
changeset
|
103 |
fun trim_blanks trees = |
2390391584c2
some document antiquotations for Isabelle/jEdit elements;
wenzelm
parents:
49650
diff
changeset
|
104 |
trees |> maps |
2390391584c2
some document antiquotations for Isabelle/jEdit elements;
wenzelm
parents:
49650
diff
changeset
|
105 |
(fn Elem (markup, body) => [Elem (markup, trim_blanks body)] |
2390391584c2
some document antiquotations for Isabelle/jEdit elements;
wenzelm
parents:
49650
diff
changeset
|
106 |
| Text s => |
61707 | 107 |
let val s' = s |> raw_explode |> trim Symbol.is_blank |> implode; |
56059
2390391584c2
some document antiquotations for Isabelle/jEdit elements;
wenzelm
parents:
49650
diff
changeset
|
108 |
in if s' = "" then [] else [Text s'] end); |
2390391584c2
some document antiquotations for Isabelle/jEdit elements;
wenzelm
parents:
49650
diff
changeset
|
109 |
|
2390391584c2
some document antiquotations for Isabelle/jEdit elements;
wenzelm
parents:
49650
diff
changeset
|
110 |
|
24264 | 111 |
|
26525 | 112 |
(** string representation **) |
113 |
||
24264 | 114 |
val header = "<?xml version=\"1.0\"?>\n"; |
115 |
||
116 |
||
26546 | 117 |
(* escaped text *) |
24264 | 118 |
|
119 |
fun decode "<" = "<" |
|
120 |
| decode ">" = ">" |
|
121 |
| decode "&" = "&" |
|
122 |
| decode "'" = "'" |
|
123 |
| decode """ = "\"" |
|
124 |
| decode c = c; |
|
125 |
||
126 |
fun encode "<" = "<" |
|
127 |
| encode ">" = ">" |
|
128 |
| encode "&" = "&" |
|
129 |
| encode "'" = "'" |
|
130 |
| encode "\"" = """ |
|
131 |
| encode c = c; |
|
132 |
||
25838 | 133 |
val text = translate_string encode; |
24264 | 134 |
|
135 |
||
136 |
(* elements *) |
|
137 |
||
26539
a0754be538ab
added output_markup (from Tools/isabelle_process.ML);
wenzelm
parents:
26525
diff
changeset
|
138 |
fun elem name atts = |
26551 | 139 |
space_implode " " (name :: map (fn (a, x) => a ^ "=\"" ^ text x ^ "\"") atts); |
24264 | 140 |
|
26525 | 141 |
fun element name atts body = |
26539
a0754be538ab
added output_markup (from Tools/isabelle_process.ML);
wenzelm
parents:
26525
diff
changeset
|
142 |
let val b = implode body in |
a0754be538ab
added output_markup (from Tools/isabelle_process.ML);
wenzelm
parents:
26525
diff
changeset
|
143 |
if b = "" then enclose "<" "/>" (elem name atts) |
a0754be538ab
added output_markup (from Tools/isabelle_process.ML);
wenzelm
parents:
26525
diff
changeset
|
144 |
else enclose "<" ">" (elem name atts) ^ b ^ enclose "</" ">" name |
24264 | 145 |
end; |
146 |
||
27884 | 147 |
fun output_markup (markup as (name, atts)) = |
38474
e498dc2eb576
uniform Markup.empty/Markup.Empty in ML and Scala;
wenzelm
parents:
38266
diff
changeset
|
148 |
if Markup.is_empty markup then Markup.no_output |
27884 | 149 |
else (enclose "<" ">" (elem name atts), enclose "</" ">" name); |
26539
a0754be538ab
added output_markup (from Tools/isabelle_process.ML);
wenzelm
parents:
26525
diff
changeset
|
150 |
|
24264 | 151 |
|
26546 | 152 |
(* output *) |
24264 | 153 |
|
43791 | 154 |
fun buffer_of depth tree = |
24264 | 155 |
let |
43791 | 156 |
fun traverse _ (Elem ((name, atts), [])) = |
26539
a0754be538ab
added output_markup (from Tools/isabelle_process.ML);
wenzelm
parents:
26525
diff
changeset
|
157 |
Buffer.add "<" #> Buffer.add (elem name atts) #> Buffer.add "/>" |
43791 | 158 |
| traverse d (Elem ((name, atts), ts)) = |
26539
a0754be538ab
added output_markup (from Tools/isabelle_process.ML);
wenzelm
parents:
26525
diff
changeset
|
159 |
Buffer.add "<" #> Buffer.add (elem name atts) #> Buffer.add ">" #> |
43791 | 160 |
traverse_body d ts #> |
26525 | 161 |
Buffer.add "</" #> Buffer.add name #> Buffer.add ">" |
43791 | 162 |
| traverse _ (Text s) = Buffer.add (text s) |
163 |
and traverse_body 0 _ = Buffer.add "..." |
|
164 |
| traverse_body d ts = fold (traverse (d - 1)) ts; |
|
165 |
in Buffer.empty |> traverse depth tree end; |
|
24264 | 166 |
|
43791 | 167 |
val string_of = Buffer.content o buffer_of ~1; |
168 |
val output = Buffer.output o buffer_of ~1; |
|
169 |
||
170 |
fun pretty depth tree = |
|
171 |
Pretty.str (Buffer.content (buffer_of (Int.max (0, depth)) tree)); |
|
25838 | 172 |
|
62663 | 173 |
val _ = |
174 |
PolyML.addPrettyPrinter (fn depth => fn _ => Pretty.to_polyml o pretty (FixedInt.toInt depth)); |
|
175 |
||
24264 | 176 |
|
177 |
||
44698 | 178 |
(** XML parsing **) |
26546 | 179 |
|
180 |
local |
|
24264 | 181 |
|
43947
9b00f09f7721
defer evaluation of Scan.message, for improved performance in the frequent situation where failure is handled later (e.g. via ||);
wenzelm
parents:
43844
diff
changeset
|
182 |
fun err msg (xs, _) = |
9b00f09f7721
defer evaluation of Scan.message, for improved performance in the frequent situation where failure is handled later (e.g. via ||);
wenzelm
parents:
43844
diff
changeset
|
183 |
fn () => "XML parsing error: " ^ msg () ^ "\nfound: " ^ quote (Symbol.beginning 100 xs); |
24264 | 184 |
|
26984
d0e098e206f3
added parse_document (optional unchecked header material);
wenzelm
parents:
26554
diff
changeset
|
185 |
fun ignored _ = []; |
d0e098e206f3
added parse_document (optional unchecked header material);
wenzelm
parents:
26554
diff
changeset
|
186 |
|
45155
3216d65d8f34
slightly more standard-conformant XML parsing (see also 94033767ef9b);
wenzelm
parents:
44809
diff
changeset
|
187 |
fun name_start_char c = Symbol.is_ascii_letter c orelse c = ":" orelse c = "_"; |
3216d65d8f34
slightly more standard-conformant XML parsing (see also 94033767ef9b);
wenzelm
parents:
44809
diff
changeset
|
188 |
fun name_char c = name_start_char c orelse Symbol.is_ascii_digit c orelse c = "-" orelse c = "."; |
3216d65d8f34
slightly more standard-conformant XML parsing (see also 94033767ef9b);
wenzelm
parents:
44809
diff
changeset
|
189 |
val parse_name = Scan.one name_start_char ::: Scan.many name_char; |
3216d65d8f34
slightly more standard-conformant XML parsing (see also 94033767ef9b);
wenzelm
parents:
44809
diff
changeset
|
190 |
|
26551 | 191 |
val blanks = Scan.many Symbol.is_blank; |
45155
3216d65d8f34
slightly more standard-conformant XML parsing (see also 94033767ef9b);
wenzelm
parents:
44809
diff
changeset
|
192 |
val special = $$ "&" ^^ (parse_name >> implode) ^^ $$ ";" >> decode; |
58854 | 193 |
val regular = Scan.one Symbol.not_eof; |
194 |
fun regular_except x = Scan.one (fn c => Symbol.not_eof c andalso c <> x); |
|
24264 | 195 |
|
26551 | 196 |
val parse_chars = Scan.repeat1 (special || regular_except "<") >> implode; |
24264 | 197 |
|
26551 | 198 |
val parse_cdata = |
199 |
Scan.this_string "<![CDATA[" |-- |
|
200 |
(Scan.repeat (Scan.unless (Scan.this_string "]]>") regular) >> implode) --| |
|
201 |
Scan.this_string "]]>"; |
|
24264 | 202 |
|
203 |
val parse_att = |
|
45155
3216d65d8f34
slightly more standard-conformant XML parsing (see also 94033767ef9b);
wenzelm
parents:
44809
diff
changeset
|
204 |
((parse_name >> implode) --| (blanks -- $$ "=" -- blanks)) -- |
26551 | 205 |
(($$ "\"" || $$ "'") :|-- (fn s => |
206 |
(Scan.repeat (special || regular_except s) >> implode) --| $$ s)); |
|
24264 | 207 |
|
26551 | 208 |
val parse_comment = |
209 |
Scan.this_string "<!--" -- |
|
210 |
Scan.repeat (Scan.unless (Scan.this_string "-->") regular) -- |
|
26984
d0e098e206f3
added parse_document (optional unchecked header material);
wenzelm
parents:
26554
diff
changeset
|
211 |
Scan.this_string "-->" >> ignored; |
24264 | 212 |
|
26551 | 213 |
val parse_processing_instruction = |
214 |
Scan.this_string "<?" -- |
|
215 |
Scan.repeat (Scan.unless (Scan.this_string "?>") regular) -- |
|
26984
d0e098e206f3
added parse_document (optional unchecked header material);
wenzelm
parents:
26554
diff
changeset
|
216 |
Scan.this_string "?>" >> ignored; |
d0e098e206f3
added parse_document (optional unchecked header material);
wenzelm
parents:
26554
diff
changeset
|
217 |
|
d0e098e206f3
added parse_document (optional unchecked header material);
wenzelm
parents:
26554
diff
changeset
|
218 |
val parse_doctype = |
d0e098e206f3
added parse_document (optional unchecked header material);
wenzelm
parents:
26554
diff
changeset
|
219 |
Scan.this_string "<!DOCTYPE" -- |
d0e098e206f3
added parse_document (optional unchecked header material);
wenzelm
parents:
26554
diff
changeset
|
220 |
Scan.repeat (Scan.unless ($$ ">") regular) -- |
d0e098e206f3
added parse_document (optional unchecked header material);
wenzelm
parents:
26554
diff
changeset
|
221 |
$$ ">" >> ignored; |
d0e098e206f3
added parse_document (optional unchecked header material);
wenzelm
parents:
26554
diff
changeset
|
222 |
|
d0e098e206f3
added parse_document (optional unchecked header material);
wenzelm
parents:
26554
diff
changeset
|
223 |
val parse_misc = |
d0e098e206f3
added parse_document (optional unchecked header material);
wenzelm
parents:
26554
diff
changeset
|
224 |
Scan.one Symbol.is_blank >> ignored || |
d0e098e206f3
added parse_document (optional unchecked header material);
wenzelm
parents:
26554
diff
changeset
|
225 |
parse_processing_instruction || |
d0e098e206f3
added parse_document (optional unchecked header material);
wenzelm
parents:
26554
diff
changeset
|
226 |
parse_comment; |
26551 | 227 |
|
228 |
val parse_optional_text = |
|
229 |
Scan.optional (parse_chars >> (single o Text)) []; |
|
24264 | 230 |
|
26546 | 231 |
in |
232 |
||
26984
d0e098e206f3
added parse_document (optional unchecked header material);
wenzelm
parents:
26554
diff
changeset
|
233 |
val parse_comments = |
d0e098e206f3
added parse_document (optional unchecked header material);
wenzelm
parents:
26554
diff
changeset
|
234 |
blanks -- Scan.repeat (parse_comment -- blanks >> K ()) >> K (); |
d0e098e206f3
added parse_document (optional unchecked header material);
wenzelm
parents:
26554
diff
changeset
|
235 |
|
40627
becf5d5187cc
renamed raw "explode" function to "raw_explode" to emphasize its meaning;
wenzelm
parents:
40131
diff
changeset
|
236 |
val parse_string = Scan.read Symbol.stopper parse_chars o raw_explode; |
26546 | 237 |
|
24264 | 238 |
fun parse_content xs = |
26551 | 239 |
(parse_optional_text @@@ |
61476 | 240 |
Scan.repeats |
26551 | 241 |
((parse_element >> single || |
242 |
parse_cdata >> (single o Text) || |
|
26984
d0e098e206f3
added parse_document (optional unchecked header material);
wenzelm
parents:
26554
diff
changeset
|
243 |
parse_processing_instruction || |
d0e098e206f3
added parse_document (optional unchecked header material);
wenzelm
parents:
26554
diff
changeset
|
244 |
parse_comment) |
61476 | 245 |
@@@ parse_optional_text)) xs |
24264 | 246 |
|
26546 | 247 |
and parse_element xs = |
43949
94033767ef9b
more precise parse_name according to XML standard;
wenzelm
parents:
43947
diff
changeset
|
248 |
($$ "<" |-- parse_name -- Scan.repeat (blanks |-- parse_att) --| blanks :-- |
94033767ef9b
more precise parse_name according to XML standard;
wenzelm
parents:
43947
diff
changeset
|
249 |
(fn (name, _) => |
43947
9b00f09f7721
defer evaluation of Scan.message, for improved performance in the frequent situation where failure is handled later (e.g. via ||);
wenzelm
parents:
43844
diff
changeset
|
250 |
!! (err (fn () => "Expected > or />")) |
43949
94033767ef9b
more precise parse_name according to XML standard;
wenzelm
parents:
43947
diff
changeset
|
251 |
($$ "/" -- $$ ">" >> ignored || |
94033767ef9b
more precise parse_name according to XML standard;
wenzelm
parents:
43947
diff
changeset
|
252 |
$$ ">" |-- parse_content --| |
94033767ef9b
more precise parse_name according to XML standard;
wenzelm
parents:
43947
diff
changeset
|
253 |
!! (err (fn () => "Expected </" ^ implode name ^ ">")) |
94033767ef9b
more precise parse_name according to XML standard;
wenzelm
parents:
43947
diff
changeset
|
254 |
($$ "<" -- $$ "/" -- Scan.this name -- blanks -- $$ ">"))) |
94033767ef9b
more precise parse_name according to XML standard;
wenzelm
parents:
43947
diff
changeset
|
255 |
>> (fn ((name, atts), body) => Elem ((implode name, atts), body))) xs; |
24264 | 256 |
|
26984
d0e098e206f3
added parse_document (optional unchecked header material);
wenzelm
parents:
26554
diff
changeset
|
257 |
val parse_document = |
d0e098e206f3
added parse_document (optional unchecked header material);
wenzelm
parents:
26554
diff
changeset
|
258 |
(Scan.repeat parse_misc -- Scan.option parse_doctype -- Scan.repeat parse_misc) |
d0e098e206f3
added parse_document (optional unchecked header material);
wenzelm
parents:
26554
diff
changeset
|
259 |
|-- parse_element; |
24264 | 260 |
|
26539
a0754be538ab
added output_markup (from Tools/isabelle_process.ML);
wenzelm
parents:
26525
diff
changeset
|
261 |
fun parse s = |
43947
9b00f09f7721
defer evaluation of Scan.message, for improved performance in the frequent situation where failure is handled later (e.g. via ||);
wenzelm
parents:
43844
diff
changeset
|
262 |
(case Scan.finite Symbol.stopper (Scan.error (!! (err (fn () => "Malformed element")) |
40627
becf5d5187cc
renamed raw "explode" function to "raw_explode" to emphasize its meaning;
wenzelm
parents:
40131
diff
changeset
|
263 |
(blanks |-- parse_document --| blanks))) (raw_explode s) of |
24264 | 264 |
(x, []) => x |
48769 | 265 |
| (_, ys) => error ("XML parsing error: unprocessed input\n" ^ Symbol.beginning 100 ys)); |
24264 | 266 |
|
267 |
end; |
|
26546 | 268 |
|
43767 | 269 |
|
270 |
||
271 |
(** XML as data representation language **) |
|
272 |
||
273 |
exception XML_ATOM of string; |
|
274 |
exception XML_BODY of tree list; |
|
275 |
||
276 |
||
277 |
structure Encode = |
|
278 |
struct |
|
279 |
||
43778
ce9189450447
more compact representation of XML data (notably sort/typ/term), using properties as vector of atomic values;
wenzelm
parents:
43768
diff
changeset
|
280 |
type 'a A = 'a -> string; |
43767 | 281 |
type 'a T = 'a -> body; |
43778
ce9189450447
more compact representation of XML data (notably sort/typ/term), using properties as vector of atomic values;
wenzelm
parents:
43768
diff
changeset
|
282 |
type 'a V = 'a -> string list * body; |
43767 | 283 |
|
284 |
||
43778
ce9189450447
more compact representation of XML data (notably sort/typ/term), using properties as vector of atomic values;
wenzelm
parents:
43768
diff
changeset
|
285 |
(* atomic values *) |
43767 | 286 |
|
287 |
fun int_atom i = signed_string_of_int i; |
|
288 |
||
289 |
fun bool_atom false = "0" |
|
290 |
| bool_atom true = "1"; |
|
291 |
||
292 |
fun unit_atom () = ""; |
|
293 |
||
294 |
||
295 |
(* structural nodes *) |
|
296 |
||
297 |
fun node ts = Elem ((":", []), ts); |
|
298 |
||
43778
ce9189450447
more compact representation of XML data (notably sort/typ/term), using properties as vector of atomic values;
wenzelm
parents:
43768
diff
changeset
|
299 |
fun vector xs = map_index (fn (i, x) => (int_atom i, x)) xs; |
ce9189450447
more compact representation of XML data (notably sort/typ/term), using properties as vector of atomic values;
wenzelm
parents:
43768
diff
changeset
|
300 |
|
ce9189450447
more compact representation of XML data (notably sort/typ/term), using properties as vector of atomic values;
wenzelm
parents:
43768
diff
changeset
|
301 |
fun tagged (tag, (xs, ts)) = Elem ((int_atom tag, vector xs), ts); |
43767 | 302 |
|
303 |
||
304 |
(* representation of standard types *) |
|
305 |
||
306 |
fun properties props = [Elem ((":", props), [])]; |
|
307 |
||
308 |
fun string "" = [] |
|
309 |
| string s = [Text s]; |
|
310 |
||
311 |
val int = string o int_atom; |
|
312 |
||
313 |
val bool = string o bool_atom; |
|
314 |
||
315 |
val unit = string o unit_atom; |
|
316 |
||
317 |
fun pair f g (x, y) = [node (f x), node (g y)]; |
|
318 |
||
319 |
fun triple f g h (x, y, z) = [node (f x), node (g y), node (h z)]; |
|
320 |
||
321 |
fun list f xs = map (node o f) xs; |
|
322 |
||
323 |
fun option _ NONE = [] |
|
324 |
| option f (SOME x) = [node (f x)]; |
|
325 |
||
47199
15ede9f1da3f
more specific notion of partiality (cf. Scala version);
wenzelm
parents:
46840
diff
changeset
|
326 |
fun variant fs x = |
15ede9f1da3f
more specific notion of partiality (cf. Scala version);
wenzelm
parents:
46840
diff
changeset
|
327 |
[tagged (the (get_index (fn f => SOME (f x) handle General.Match => NONE) fs))]; |
43767 | 328 |
|
26546 | 329 |
end; |
43767 | 330 |
|
331 |
||
332 |
structure Decode = |
|
333 |
struct |
|
334 |
||
43778
ce9189450447
more compact representation of XML data (notably sort/typ/term), using properties as vector of atomic values;
wenzelm
parents:
43768
diff
changeset
|
335 |
type 'a A = string -> 'a; |
43767 | 336 |
type 'a T = body -> 'a; |
43778
ce9189450447
more compact representation of XML data (notably sort/typ/term), using properties as vector of atomic values;
wenzelm
parents:
43768
diff
changeset
|
337 |
type 'a V = string list * body -> 'a; |
43767 | 338 |
|
339 |
||
43778
ce9189450447
more compact representation of XML data (notably sort/typ/term), using properties as vector of atomic values;
wenzelm
parents:
43768
diff
changeset
|
340 |
(* atomic values *) |
43767 | 341 |
|
342 |
fun int_atom s = |
|
43797
fad7758421bf
more precise integer Markup.properties/XML.attributes: disallow ML-style ~ minus;
wenzelm
parents:
43791
diff
changeset
|
343 |
Markup.parse_int s |
fad7758421bf
more precise integer Markup.properties/XML.attributes: disallow ML-style ~ minus;
wenzelm
parents:
43791
diff
changeset
|
344 |
handle Fail _ => raise XML_ATOM s; |
43767 | 345 |
|
346 |
fun bool_atom "0" = false |
|
347 |
| bool_atom "1" = true |
|
348 |
| bool_atom s = raise XML_ATOM s; |
|
349 |
||
350 |
fun unit_atom "" = () |
|
351 |
| unit_atom s = raise XML_ATOM s; |
|
352 |
||
353 |
||
354 |
(* structural nodes *) |
|
355 |
||
356 |
fun node (Elem ((":", []), ts)) = ts |
|
357 |
| node t = raise XML_BODY [t]; |
|
358 |
||
43783 | 359 |
fun vector atts = |
46839
f7232c078fa5
simplified -- plain map_index is sufficient (pointed out by Enrico Tassi);
wenzelm
parents:
46837
diff
changeset
|
360 |
map_index (fn (i, (a, x)) => if int_atom a = i then x else raise XML_ATOM a) atts; |
43778
ce9189450447
more compact representation of XML data (notably sort/typ/term), using properties as vector of atomic values;
wenzelm
parents:
43768
diff
changeset
|
361 |
|
43844 | 362 |
fun tagged (Elem ((name, atts), ts)) = (int_atom name, (vector atts, ts)) |
43767 | 363 |
| tagged t = raise XML_BODY [t]; |
364 |
||
365 |
||
366 |
(* representation of standard types *) |
|
367 |
||
368 |
fun properties [Elem ((":", props), [])] = props |
|
369 |
| properties ts = raise XML_BODY ts; |
|
370 |
||
371 |
fun string [] = "" |
|
372 |
| string [Text s] = s |
|
373 |
| string ts = raise XML_BODY ts; |
|
374 |
||
375 |
val int = int_atom o string; |
|
376 |
||
377 |
val bool = bool_atom o string; |
|
378 |
||
379 |
val unit = unit_atom o string; |
|
380 |
||
381 |
fun pair f g [t1, t2] = (f (node t1), g (node t2)) |
|
382 |
| pair _ _ ts = raise XML_BODY ts; |
|
383 |
||
384 |
fun triple f g h [t1, t2, t3] = (f (node t1), g (node t2), h (node t3)) |
|
385 |
| triple _ _ _ ts = raise XML_BODY ts; |
|
386 |
||
387 |
fun list f ts = map (f o node) ts; |
|
388 |
||
389 |
fun option _ [] = NONE |
|
390 |
| option f [t] = SOME (f (node t)) |
|
391 |
| option _ ts = raise XML_BODY ts; |
|
392 |
||
43768 | 393 |
fun variant fs [t] = |
394 |
let |
|
43778
ce9189450447
more compact representation of XML data (notably sort/typ/term), using properties as vector of atomic values;
wenzelm
parents:
43768
diff
changeset
|
395 |
val (tag, (xs, ts)) = tagged t; |
43768 | 396 |
val f = nth fs tag handle General.Subscript => raise XML_BODY [t]; |
43778
ce9189450447
more compact representation of XML data (notably sort/typ/term), using properties as vector of atomic values;
wenzelm
parents:
43768
diff
changeset
|
397 |
in f (xs, ts) end |
43767 | 398 |
| variant _ ts = raise XML_BODY ts; |
399 |
||
400 |
end; |
|
401 |
||
402 |
end; |